What determines which genes are activated by a transcription factor (TF), which are repressed, and which are not regulated? The simplest conceptual model is that a TF directly regulates the genes in whose regulatory regions it binds. In this ?independent action model?, if the TF has an activation domain, it activates the genes whose promoters it binds; if it has a repression domain, it represses the genes whose promoters it binds. It is remarkable then, that in existing, comprehensive data sets, that the vast majority of genes whose promoters are bound by a TF show no change in expression when the TF is knocked down or knocked out. In yeast data where all TFs have been analyzed, only 3-5% of bound genes respond when the TF is knocked out; in more recent, but less comprehensive human data it is rarely more than 15%. Furthermore, nearly every TF appears to activate some of the genes it binds while repressing others. These data show that we do not have a credible rough draft map of the regulatory logic of any eukaryote, as most of the genes bound by a TF in existing data are not regulated by that TF and we do not know how to predict which ones are. Our preliminary results show that some, but clearly not all, of the problem can be explained by technical issues inherent to the existing datasets. We there propose to generate new, comprehensive data sets that are ideally suited to studying the relationship between binding and functional regulation (Aim 1). We will use these datasets to quantify the predictive power of the independent action model. In Aim 2, we will collect genome- wide data focused understanding the interactions between TF binding and local chromatin state to see we can use the information to extend the independent action model to increase its predictive power. Finally, we will collect genome-wide data to identify TF-TF interactions that affect the gene expression response to perturbation. We will assess the predictive power of this information using computational models. This project will produce the first consistent data set that is well suited to studying the relationship between TF binding and regulation as well as the first credible overview of regulatory logic for any eukaryote. At the conclusion, we will know why perturbation of a TF apparently activates some of the genes it binds, represses other genes it binds, and does nothing at all to most of the genes it binds. Is it just the limitations of existing high-throughput data sets, or do interactions dominate independent action?